Laura E. Boeschoten, Joannès Guillemot, Juliano van Melis, Lourens Poorter, Ricardo R. Rodrigues, Frans Bongers, Paulo G. Molin, Marielos Peña-Claros, Vinicius C. Souza, Cássio A. P. Toledo, Patrick Faria Fernandes, Marcelo P. Ferreira, Angélica F. Resende, Laura H. P. Simões, Catherine Torres de Almeida, María Uriarte, Pedro H. S. Brancalion
Abstract
Tropical landscapes are undergoing rapid transformation due to human activities and global change. Forest restoration has emerged as a key strategy to mitigate biodiversity loss and climate warming. However, a standardized assessment of how different restoration methods contribute to biodiversity recovery and conservation remains lacking. Here, we present the first comprehensive comparison of tree diversity restoration and the drivers of recovery across five main reforestation methods (naturally regenerating forests, biodiverse restoration plantings, short- and long-rotation tree monocultures, agroforests) relative to three reference systems (agropastoral lands, degraded and conserved forest remnants). Tree inventories were conducted in 519 plots (900 m2 each) across two forest types (rainforest and seasonally dry forest) in the Atlantic forest of São Paulo state, Brazil, encompassing over 39,000 trees and 869 species. We found that: (1) all reforestation methods except short-rotation monoculture plots supported tree diversity recovery. In the rainforest, conserved remnant plots maintained the highest average Shannon diversity (Hill 1 = 29 ± 14), while naturally regenerating forests and restoration plantings approached the diversity of degraded remnant plots (15 ± 5). In seasonally dry forest, biodiverse restoration plantings and agroforests reached diversity levels comparable to conserved remnants (15 ± 6). Additionally, (2) recovery was influenced by forest age, climate (water availability), soil fertility, and landscape context, though the relative importance of these factors varied by method. Climate and landscape context were more influential for recovery in naturally regenerating forests, while soil conditions played a greater role in biodiverse restoration plantings. Lastly, (3) species composition in naturally regenerating forests most closely resembled that of conserved remnants. Conversely, restoration plantings and agroforests exhibited high compositional overlap across sites, reducing overall species richness. Our findings underscore the wide variation in biodiversity outcomes among and within reforestation methods, emphasizing that goals and strategies must align with local conditions to maximize benefits in complex tropical landscapes.
